TheRadioBoard

Hi eric,

By not quite as good as a superhet I see two possibilities

1) The bandwidth of a superhet on MW is constant (ideally 8 to 9 khz) regardless of whether it is tuned to the low or high end of the band, whereas a regenerative receivers bandwidth will vary with tuning frquency assuming the same level of regeneration is applied - I dont think this is too important as one can still have an average bandwidth that is very usable

2) For fixed or automatic regeneration as one of the goals we may not be able to obtain the maximum sensitivity/selectivity one could obtain from a regenerative receiver that was set up carefully by manual operation.

Gregcon

Theoretically, at resonance the impedance of a parallel-resonant circuit is infinite. But I understand what you're after. At resonance, the reactances of inductor and capacitor are equal but of opposite signs. If your 140-pF cap tunes to 540 kHz, its reactance at that frequency is 2105 ohms. The inductance is 620.5 uH. Maintaining that inductance, the capacitance to resonate at 1650 kHz is 15 pF, and the L and C reactances are 6431 ohms.

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

Hi Jim,

From your calculations does this mean if input impedance was say 100k then this would not have an effect on the q of the tuned circuit as this would be 6431 ohms in parallel with 100k? or have I got it wrong?

Gregcon

@Ham-er:


@Gregcon:



I don't think that such a regen could be better than a superhet. The sensivity will be ok, but the skirt-selectivity will always suffer with only one tank circuit. I think that a regen stage with 3 transistors will suffer from instabilities like backlash, howling at audio frequencies, etc, because of the high open loop gain. At higher frequencies in the shortwave bands these 3 transistors will introduce unwanted fase shifts, making the right amount of regeneration difficult.

We're only talking about the tuned circuit, apart from its use in a regenerative stage. As the circuit nears oscillation, signal level increases, as does apparent Q. If Q and signal level are increasing, so must the impedance of the tuned circuit. This is the advantage of a regenerative stage: Better Q and lots of amplification.

The input is usually not coupled directly across the tuned circuit. As the input impedance is usually low, either a separate winding of fewer turns, a tap on the winding, or some form of capacitive coupling is used. It would be hard to match a common antenna input up to 100k ohms in any event. As the regenerative stage has so much gain, you don't need much coupling. Your interest in reducing loading of the tuned circuit is good, but I still wouldn't hang 100k across it.

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

Hi Drm,

I tend to agree with you.

With the circuit I built I did not have problems with audio howling as I only had RF feedback and the oscillation point seemed quite constant for the entire MW band. However I did have breakthrough from powerful local stations (2km away) and whatever I tried (except a second tuned circuit which I did not want to do) I could not get rid of this.

Does anyone know how I could possibly overcome this without too much complexity or reducing sensitivity? or is this the best one can have with a single tuned regen?

gregcon

Hi Jim again

Thanks, I got all that and I suspected that a much higher input impedance was needed. When browsing through similar circuits i noticed that either a tapping or smaller coupling coil was used, or the input impedance was in the megohms (as in the prototype i built)

gregcon

Hi DrM,

This is exactly why I was asking Gregcon to define what he meant by "performance".

Skirt selectivity and stability ARE some of the things that are "better" in a superhet than a regen.

73
kb0lxy

Coming back to this again -- Wherever a circuit parameter can vary as a function of its operation, it's best to keep the impedance as low as possible. A 1% variation in, say, loading, at 100K is 1000 ohms, whereas at 50 ohms it's only 0.5 ohm. Maintaining good impedance matching is easier at lower impedances.

Also, P = E^2 / Z. (Voltage = SQRT P x Z) So for a given power level, voltage increases quickly with impedance. (You can run some numbers here and see: http://www.csgnetwork.com/ohmslaw2.html.) A negative side effect is that it gets harder to keep signals where they belong when dealing with high-impedance circuits. Sometimes we have no choice. For example, the impedance of a regenerative stage tank circuit is very, very high. But where there is a choice, designing for a lower impedance, especially when coupling one stage to another, is a better option.

(This points back to one of the problems with regenerative detectors on the MW broadcast band. Across the >3:1 frequency range, LC reactances increase, as does tank voltage. While this improves gain at higher frequencies, it also makes it harder to obtain smooth regeneration control as you tune up the band. National Radio, in the SW-3, and KR1S in a solid-state design, partially solved this problem by breaking the band into segments, to better achieve usable LC ratios.)

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

Jim
Thanks for overview and history - Many good points.
You are right - The most logical path would be a regenerative superhet but it would be nice to see the maximum one could get from a pure regenerative.

Many years ago, an author for "Radio constructor" described circuits based on the Super alpha regenerative receiver. The link is

http://www.radioconstructors.info/sdh/sdh.html

A good circuit description is given in the SA4 article although later he brought out improved versions.

He claims he has done basically what I am asking (i'm sceptical)
ie Good selectivity, even with strong local stations on nearby frequencies and no adjustments of regeneration over the entire MW band. He further compares a superhet to his circuit and says only on strong local stations was the selectivity slightly better.

From what I gather the extra performance claimed is due to a high L/C ratio in the tuned circuit - he uses 50pf fixed with a variable ferrite inductor and a very high impedance offered by two common collector stages, In the article he says that the detector diode on the output of the second stage somehow play a role in greatly improving skirt selectivity though I could no follow him on this

Maybe there is something in this - my own view is that to achieve the best skirt selectivity from a regenerative one would need a superhigh Q tuned circuit to be totally unloaded by the circuit and an rf amplifier which would provide most of its gain by regeneration and have limited gain when no regeneration is present - is this right?

Gregcon

Message removed to a new topic.

The Sir Doug radio is a reflex, not a regen. Since the 1930s, regen builders have avoided high LC ratios. There are ways to avoid regen adjustment without going to high LC ratios-- see vladn's posts in Tube Radios -- but they add back circuit complexity.

If you want a regen's benefits, having to tweak regeneration doesn't seem like a big sacrifice.

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

Jim
The author claimed that the high LC ratio contributed to extra selectivity and sensitivity somehow although I cant understand the logic. The input circuit had a very low input capacitance which in turn meant he could use a lower cap value of 50pf for tuning the entire MW band. The reflex circuit does however add unnecessary complexity to optimising the design, as audio gain can easily be achieved with an extra stage

I wonder if anyone out there has built a spontaflex super-alpha MW receiver and whether the performance is as claimed

gregcon

Use the formula Q = Rp * sqrt(C/L) for an estimate of the required input impedance at the boundaries of the band. Ferrite rod antenna/tank can achieve Q of few hundreds (without regeneration). Expressing Rp from the formula above you get Rp = Q / sqrt(C/L), assuming Q around 300 and substituting C and L values from you circuit you can get the ballpark estimate of the intrinsic tank Rp (it is approximate since in practice Q depends on f). When the load Rp is around or below this value the intrinsic Q of the resonant circuit will start dropping.

If you do not have the value of L you can use the frequency instead. Skipping the derivation the formula is:
Rp = Q/(C*2*pi*f)

BTW the input impedance of your input RF buffer will be likely limited by the Miller capacitance. In your case the gain is small (which is good). But still a cascode with jfet at the bottom will give you the highest input impedance and stability (and high gain as a bonus).

Since you are using essentially the amplified AGC signal to control the gain in the regeneration loop you may want to read the following link, the guy built an automatic regen control this way:
http://cool386.tripod.com/arc/automatic_regen.html

I think the breakthrough problem is an inevitable evil of any single tuned regen. In fact AGC makes it more noticeable. Your options are (i) double tuned design, (ii) fixed rejection filter(s) for the strongest local(s), or (iii) superhet

That's very interesting. About a year ago I was scouring the net for all resources related to automatic regeneration control, but didn't see this. I notice the date is fairly recent, January 2012.

I had sometimes thought that a tracking BFO, that tracks the regen frequency, would instantly provide a shrieking tone if the regen was adjusted above the oscillation threshold. This above-threshold indicator might be usable as some sort of an automatic regeneration control. It seems the author of the above approach managed to do it without a separate BFO by causing the regen itself to squeal above threshold. Very neat.

Doing another web search just right now on automatic regeneration control, I notice another circuit that didn't seem to exist last time I searched one year ago. See circuit #5 here: http://zpostbox.ru/r1_e.htm . Any comments on that circuit?